NSCT-Based Infrared Image Enhancement Method for Rotating Machinery Fault Diagnosis

Infrared images are usually subject to low contrast, edge blurring, and high noise. Especially for machinery diagnosis, the range of temperature variation is narrow, which causes the difficulty to diagnose different equipment conditions directly from infrared images. To enhance fault-related information extraction and improve diagnosis accuracy, a new infrared image analysis method based on nonsubsampled contourlet transform is investigated with fuzzy enhancement and nonlinear gain. The parameters of fuzzy enhancement and nonlinear gain functions are optimized by particle swarming optimization algorithm, in which the optimization criterion is formulated by information entropy and contrast of infrared image. Feature extraction and dimensionality reduction methods are then applied to select features for further diagnosis. The effectiveness of the presented method is experimentally validated in the infrared image analysis of rotor test stand in laboratory, and the results show that the presented method can effectively enhance the fault information, enlarge the contrast of image under different conditions, and improve the accuracy of machinery fault diagnosis.